The following content is translated from Chinese source by machine translation without post-editing.
In this article, we will provide two relevant project cases to illustrate the new language related needs of customers in the era of AI and how TalkingChina Translation develops and delivers solutions to these needs. With the advent of the era of artificial intelligence, more and more language service demands are no longer appearing in traditional forms, which puts higher demands on translation companies: we need to have the ability to provide customized services, focus on customer needs, take our rich global multilingual native translator resources, multilingual delivery capabilities, customer communication and personalized solution development capabilities, and project comprehensive management capabilities as core competitiveness, to meet the diverse new language service needs of customers in the process of globalization.
Case 1
Project Background
The client company is a leading AI technology service company. By utilizing technologies such as big language modeling, natural language processing (NLP), deep machine learning, privacy computing, and cloud computing, we provide services through Model as a Service (MaaS) and Business as a Service (BaaS) service models. Our products and services are mainly applied in various industries such as banking, consumer goods, insurance, e-commerce, automotive, logistics, ticketing, energy, and construction. The customer requires native speakers of multiple languages to record audio for training an intelligent voice robot focused on promoting and urging bank financial products, achieving digitization of sales and customer service. The customer hopes to enhance the robot’s interaction ability through these real voice samples, making it more vivid and natural when communicating with the target customer group.
Customer requirements
1. This project requires three different native English speakers with different accents from different regions (London English, American English with a Washington Northern accent, Singaporean English) as well as native Spanish speakers from Mexico to record.
2. Participants will record based on the recorded text provided by the client, and the recording device can be a mobile phone. Participants are not required to have a professional dubbing background. However, due to the relatively written nature of the provided text, the client hopes that the voice actors can respond flexibly and transform the written content into a more colloquial and emotionally appropriate expression based on the characteristics of different roles.
3. The project mainly includes two aspects of language service requirements:
3.1 Review of recorded text. It is necessary to make modifications to the text in terms of language and adaptability to the oral expression of recordings;
3.2 Record according to the scene requirements, and the recording will include two characters: AI character and User character.
Project difficulties
1. Difficulty in finding resources: Regional restrictions are very strict, requiring not only a balanced gender ratio of voice actors, but also ensuring that their voices and voice emotions meet the specific requirements of clients;
2. High requirements for project management capabilities of translation companies: As this is an unconventional translation project, some resources lack relevant work background. Therefore, project management personnel need to develop relevant guidelines and training resources based on customer requirements and feedback to facilitate training and help more resources improve project delivery capabilities, expand resource scope, and reduce dependence on some mature resources;
3. The quotation method is based on hourly rates, and the client will provide an approximate range of working hours that is acceptable within a reasonable range. However, the unit price is low, so the translation company can only make the greatest effort in the project management “impossible triangle” of price, quality, and time.
TalkingChina’s Translation Response Plan
How to face resource challenges:
Considering the effectiveness of the subsequent recording, we have adopted a work approach where the same person is responsible for both recording text review and recording. This choice not only helps improve the accuracy of proofreading, but also lays a good foundation for subsequent recording effects.
In response to project requirements, we actively seek local native speakers with call center and telemarketing company backgrounds in specific regions through social media software.
1.In the process of resource screening, we require the client to provide a sample text so that we can send it to interested individuals for audio testing. At the same time, we conducted online meetings with clients to thoroughly communicate details such as voice tone and intonation, in order to ensure accurate understanding of their needs and expectations. After preliminary screening, we recommend excellent audio samples to our clients. After the customer confirms, we will proceed with the proofreading of the recorded text.
2. Execution of audio text proofreading work: As audio texts are presented in a conversational form, we emphasize the use of colloquial expressions during the proofreading process, avoiding complex long sentences and prioritizing short and clear sentences to better convey information. In addition, the language used should be as close as possible to the most commonly used expressions of the local people, which not only enhances the affinity of the text but also ensures the understanding of the audience. However, while pursuing colloquialism, we will also strictly control to ensure that the original meaning of the sentence is not changed.
3. Execution of recording work: We guide participants to record in a lively and infectious way, avoiding rote memorization and creating a real atmosphere for dialogue. During the recording process, the background sound effects need to be consistent to ensure overall coherence. There are two roles in the recording: AI role and User role. We guide project participants to demonstrate natural, enthusiastic, friendly, and persuasive qualities when recording AI characters, while the User character should be as close to the daily state of answering phone calls as possible. The overall tone should avoid being too plain or dignified, and the recording personnel can imagine themselves answering a marketing call for a relaxed interaction, the more natural the better. It should be noted that the finalized text cannot be modified during recording, but mood words such as “OK”, “Emmm”, “Sure”, “Wow” can be freely added to enhance the sense of liveliness.
4. Continuous training for recording personnel: Before the official recording begins, we have conducted sufficient communication and online training for participants to ensure that they can master the tone and state of the recording. After completing the recording of the first project, we will further communicate and train the recording personnel based on customer feedback. This process aims to help them quickly find the recording tone state that meets the requirements, which can better simulate real conversation scenes. All training and guidance materials accumulate into the knowledge assets of the project, forming a complete set of audio samples and written standardized requirements.
5. Carry out sufficient warning work:
Firstly, we signed an agreement with each participant to waive their personal voice intellectual property rights, and also clearly agreed with the client on the purpose of recording the voice to avoid potential disputes in the future.
Secondly, subtle differences in tone, emotional expression, and other aspects of speech may result in some degree of rework. Therefore, before the project begins, we need to reach an agreement with all project participants and clients to clarify under which circumstances re recording can be done for free and under which circumstances additional fees will be charged. This clear agreement not only helps to control project costs, but also ensures that the project progresses in an orderly manner according to the established schedule and quality standards, avoiding disputes.
Case 2
Project Background
The client company is a technology enterprise focused on intelligent vehicle innovation, dedicated to the fields of new energy power, intelligent cockpit, and intelligent chassis, deeply empowering traditional vehicles to achieve intelligent upgrades. In order to improve the performance of its in car voice system, the customer has put forward multiple requirements, including instruction expansion, multilingual instruction, and instruction recording for native speakers. By collecting these authentic voice samples, customers hope to enhance the interactive capabilities of the voice system, enabling it to accurately and effectively recognize user voice commands.
Customer requirements
1. Expansion and Multilingualization of Instructions
The customer provided all Chinese functions in their in car voice system. For each Chinese function, we will expand at least 20 related voice commands based on its specific purpose. These instructions need to be in line with daily usage scenarios and expressed in a colloquial manner to ensure that users can easily and naturally interact with the system in future practical use.
For example:
Primary function: Air conditioning module
Secondary function: Turn on the air conditioning
According to the secondary function, at least 20 instructions need to be expanded
Languages involved: English, Russian, Arabic.
Native language recording requirements
Russian native speakers and Arabic native speakers from the United Arab Emirates are required to make separate recordings based on previous multilingual instructions. When recording, it is necessary to speak English and Russian, English and Arabic naturally and fluently.
The client needs to record audio in both English and Russian, as well as English and Arabic, according to the specified text at designated venues in Wuhan and Shanghai. Each language requires 10 recorders (5 males and 5 females), and the recording scenes include the client’s office and a real car on the road. The audio content requires accuracy, completeness, and fluency.
Project difficulties
Limited budget;
There are high requirements for the project management ability of translation companies: the expansion and multilingual of instructions are unconventional projects that require project management personnel to develop relevant guidelines and training resources based on specific customer requirements and feedback;
Resource scarcity: The client requests that Arabic language recordings must be conducted by native speakers from the United Arab Emirates and must be recorded on-site in a designated city, while also considering the gender ratio of the recording personnel. Arabic native speakers from other countries are not accepted.
TalkingChina’s Translation Response Plan
How to deal with resource challenges:
1.1 Based on the characteristics of the project, we first choose to complete the English expansion of the instruction. We searched for native English speakers with high cooperation, quick feedback, and a deep understanding of project requirements in TalkingChina’s vast resource library. We prioritized expanding the 20 instructions and sent them to the client for confirmation. Based on customer feedback, we continuously update the guidelines and provide ongoing training. During the expansion process, we always maintain close cooperation with customers and promptly raise questions about functional points in order to better understand their needs. We will use English as a template to expand the instructions in Russian and Arabic. This strategy not only improves work efficiency and reduces rework rates, but also lays a good foundation for subsequent recording effects.
1.2 In response to the recording project requirements, we have started searching for native speakers in Wuhan, Shanghai, and surrounding areas. As a result, Russian native language resources were quickly available, but domestic resources for Arabic were extremely limited and the costs generally exceeded the client’s budget. In this situation, we have had multiple communications with the client regarding their demand for Arabic language recording, and ultimately reached a compromise solution: introducing remote recording by overseas Emirati during static recording in the office; During the dynamic recording of the actual vehicle on the road, some Arab native speakers from non UAE regions are invited to participate in the project.
2. Execution of recording work: When conducting offline recording, we have prepared a detailed English recording requirements guide in advance for each native speaker participating in the recording, and developed a detailed schedule based on the time of the client and participants. For remote recording, we also provide English recording guidance for each project participant and arrange work according to the client’s needs. Participants are required to record using a computer or mobile phone in a quiet environment, maintaining a distance of 20 to 40 centimeters from the device to simulate the interaction between the car and the onboard system, and have conversations at normal volume. Before the official recording begins, we require each participant to record a sample sound and wait for the client’s confirmation before starting the official recording.
Project Summary and Prospect
With the rapid development of AI technology, the language service industry is undergoing unprecedented changes. The continuous emergence of new language service demands poses new challenges to the service models and capabilities of translation companies. Here is a summary of these two cases and some thoughts on the future:
1. Innovative service models: Traditional language services can no longer meet the current market demand, and innovative service models such as Model as a Service (MaaS) and Business as a Service (BaaS) are becoming new industry standards. TalkingChina Translation Company demonstrated how to flexibly adapt and lead this change through these two cases.
2. The combination of technology and humanities: In the era of AI, the advancement of technology is combined with the subtle emotions and cultural backgrounds of human language to provide customers with a more personalized and natural service experience. Our project practice has shown that by deeply understanding customer needs and cultural differences, we can create more attractive language service products.
3. New challenges in project management: Project management is no longer just about managing processes and time, but more about managing talent, technology, and innovative thinking. TalkingChina Translation Company demonstrated through these two cases how to overcome budget and resource constraints through refined management and innovative thinking.
4. Global resource integration: In the context of globalization, the demand for language services transcends national borders, requiring translation companies to have the ability to integrate global resources. Our case demonstrates how to find and utilize the most suitable resources through a global network to meet specific customer needs.
5. Intellectual property protection: While providing language services, we must also attach importance to the protection of intellectual property rights. By signing agreements with participants to clarify the scope and purpose of the use of voice materials, we can prevent potential legal risks.
6. Continuous learning and adaptation: With the rapid development of AI technology, translation companies must continuously learn the latest technology and industry trends to adapt to the constantly changing market environment. Our project experience has accumulated valuable knowledge assets, laying a solid foundation for future development.
Both of these cases have achieved good results so far. We not only provide high-quality language services to our clients, but also set new service standards for the industry. In the future, TalkingChina Translation Company will continue to explore and innovate to meet the constantly evolving language service needs in the AI era.
Post time: Jun-09-2025